Interesting, sbaker. I downloaded the .pdf for future reference.
Referring to the paper of footnote 7, Walterelt et al. state that the natural growth direction on currently available substrates yields an undesirable orientation wrt electrostatic fields within the epitaxial structure. The substrates discussed in that paper appear to be sapphire and silica glass. In their review of the art, they state that on sapphire and on silica glass, researchers have only been successful in growing GaN along the [0001] direction, i.e. normal to the plane represented similarly (if I remember this stuff correctly). The substrate 6H-SiC(0001) is also discussed. In Figure 2a, they show C-plane GaN(0001) grown on 6H-SiC(0001), the Si-face.
The authors have grown M-plane GaN(1100) on tetragonal LiAlO2(100) with a mismatch of only 1-2%. The multiple quantum well structures fabricated with this crystal orientation are practically free of internal electrostatic fields. This has the effect of increasing overlap of the electron and hole wavefunctions within the quantum well and thereby reducing radiative lifetimes (inversely proportional to overlap integral) by a factor of 10-100. The result of all this is greater quantum efficiency, since energy losses to competing, non-radiative processes are lower due to the shorter lifetime.
{just trying to whittle it down}
This paragraph from the article gives some symmetry arguments relating to electrostatic polarization in this type of crystal and makes clear what the authors are trying to do:
A material is polarized at equilibrium if it has a singular polar axis: this is the case with the wurtzite structure, with its [0001] axis. The zinc blende structure, in contrast, cannot be polarized as it has four symmetry-equivalent polar axes, whose contributions cancel each other. The very uniqueness of the singular [0001] polar axis of the wurtzite phase implies that any direction orthogonal to it, such as [1100] (ref.15), will not carry spontaneous polarization components. At the same time these directions will carry no piezoelectric polarization if shear stresses in the growth plane are absent, as is the case in our present study...
They also state that no suitable, stable substrate with the zinc blende structure is available. Of course, if a GaN substrate were available in either crystal structure, epitaxial growth of GaN/(Al,Ga)N in a non-polar direction could be easily done with no mismatch.
With LiAlO2, there is the question of whether or not it is or can be made conducting.
WT |
